bartlomieju / rslint Goto Github PK

Unbeatably fast
Highly parallelized (files linted in parallel, rules run in parallel, nodes could be traversed in parallel in the future)
Rich, cross-platform, colored diagnostics with secondary labels, primary labels, and notes
Lossless untyped node and token driven linting allowing easy traversal of the syntax tree from any node
Automatic docgen for rule documentation removing the need for writing rustdoc docs and user facing docs
Distinctly grouped rules
Rule examples generated from tests
Easy macros for generating rule declarations and config fields
No need for dealing with script/module or ecma versions, linter deduces source type and assumes latest syntax
No need for a configuration file
Completely error tolerant and fast parser
Lossless tree used for stylistic linting
TOML config (json will be allowed too), (TOML implemented, json not yet)
Incremental reparsing and native file watching support (WIP, see #16)

Planned

Global config
SSR-like templates for node matching and autofix
Autofix without requiring reruns of all rules
WASM builds

Speed

RSLint is designed to be the fastest JavaScript linter ever made, it accomplishes this in various ways:

Using a custom fast parser which retains whitespace
Using a lookup table and trie based lexer for parsing
Using separate distinct threads for splitting up IO bound tasks such as loading files
Linting each file in parallel
Running each rule from every group in parallel over the concrete syntax tree
(WIP) linting each untyped node in parallel
(WIP) Incrementaly reparsing and relinting files
(WIP) Having native file watching support using incremental parsing

RSLint's goal is to provide extremely fast and user friendly linting for the whole js ecosystem. There are tons of things to do to bring it up to par with existing linters. This is a list of planned features and things to do ranked in order of highest to lowest priority (this is by no definition final, things will change):